A fuzzy decision variables framework based on directed sampling for large-scale multiobjective optimization

Abstract

The enormous search space in large-scale multiobjective optimization presents a significant challenge to the convergence of existing evolutionary algorithms (EAs). It is necessary to further improve the convergence efficiency of the fuzzy decision variables framework (FDV) for large-scale multiobjective optimization. Therefore, this paper proposes a fuzzy decision variables framework based on directed sampling (FDVDS) for large-scale multiobjective optimization. After initializing the population, guided solutions are generated by directed sampling to improve the diversity of the population and speed up the convergence of the population. Finally, some representative EAs are embedded into FDVDS, and the framework’s effectiveness is verified by performing comparative experiments on large-scale multiobjective optimization test problems with up to 5000 decision variables.

Description

Keywords

Directed sampling, evolutionary algorithms, large-scale optimization, multiobjective optimization

Citation

Wang, S., Zheng, J., Zou, J., Liu, Y., Yang, S. and Zou, Y. (2023) A fuzzy decision variables framework based on directed sampling for large-scale multiobjective optimization. In Genetic and Evolutionary Computation Conference Companion (GECCO ’23 Compan ion), July 15–19, 2023, Lisbon, Portugal, pp. 419-422

Rights

Research Institute

Institute of Artificial Intelligence (IAI)